Figma adds tools to guide AI agents and design systems
At its first Config event in India, Figma introduced a Markdown-based editor called Figma Specs that lets teams define design language and give both people and AI agents a shared reference. The company also moved its code layers feature into public beta, enabling code within canvas designs and letting an agent turn designs into code or modify it. Figma says these additions are meant to make AI-assisted design more predictable and aligned with brand rules.
Figma unveiled Specs at its inaugural Config India gathering. The Markdown editor lets teams document design language, either as a separate file or alongside the canvas, and link to canvas items, FigJam diagrams, code layers, and designs. It can also incorporate design and code libraries plus guidelines, giving agents a central reference.
The code layers feature, first shown in private beta in June, is now in public beta. It embeds code in canvas designs for interactive prototypes, and an agent can convert designs to code or edit code. Figma also supports over 20 MCP tools, including Open Code, Dia, Grok Bot, and Sarvam, and integrates Payload with Figma Make.
These tools could affect designers, developers, product teams, and brands by making AI-assisted work more consistent with established design rules. Teams may spend less time correcting agent output, while smaller organizations could adopt shared design references more easily. However, reliance on centralized specs and code-generation agents may raise questions about oversight, skill shifts, and whether brand guidelines remain adaptable. The impact will likely depend on how transparently these systems handle errors and context.